6 Market Research

Descriptive research methods 

• Observation 

• interviews/focus groups 

• Surveys 

• Conjoint analysis 

• Brain imaging/physiological tracking

• Social listening

Causal research methods 

• Experiments 

• Statistical/econometric methods

  • Instrumental variables, natural experiments

Observation

• Observing usage behaviors is great for discovering consumers’ unique innovations

• Observation at scale 

  • RFID tags, phones, cameras help track the movement of customers in retail environment 

  • Technology powered by advancements in camera technology, decreasing costs of cameras

Surveys 

1. Avoid question ambiguity/confusion

2. Avoid double-barreled questions

3. Avoid leading questions

Sampling/response bias

• Hard to get perfectly representative samples

• The individuals who respond are often self selected

• Keep in mind how sample was collected

  • How sample affects participation 

  • Aim for high response rates

  • Re-weight results based on known demographics 

Projective survey techniques 

• Tap into consumers deep attitudes, motivations, beliefs 

• Allows marketers to capture emotions and associations that lie below conscious awareness

• Psychodrama technique 

  • Trying to uncover deep emotion drives in decision making—Is Dove soap masculine or feminine?

• ZMET technique 

  • Studying obvious associations of your brand

    • Nestle revealed their brand is a reminder of pleasant childhood memories

• IAT test

• Implicit Associations Test

• Measures errors and reaction time

• Common issues 

  • Low test-retest reliability

  • Weak evidence of link between IAT scores and behavior 

  • Could IAT be measuring familiarity with an association?

• Lots of stimuli in questions

Common survey issues

• People lie 

• People don’t know why they like what they like 

• The questions shape the answers

• fMRI can tell us what self-report can’t 

• But neural precision can receive too much credit

  • Solution: Add neuroscience jargon, and people are more likely to believe it

Emerging technologies: people tracking 

• RFID tags, mobile devices, and cameras can help track the movement of customers in retail environments 

• Concerns

  • Loss of privacy 

  • Racial bias 

    • Facial recognition performs worse on non-white, non-males

  
  

  • Legislation 

    • Some locations banned usage

Correlation =/ causality 

• Correlation: movement in one variable is related to movement in another variable

• Causation: movement in one variable produces movement in another variable

• Variables:

  • X = Independent variables (factors to manipulate) 

  • Y = Dependent variable (behaviors, perceptions, feelings) 

  • Z = Confounds that affect both X and Y

3 factors to draw casual conclusions 

• Correlation 

• No third factor 

• Temporal antecedence 

  • The cause must precede the effect (order of events)

Experiments

• Experiments are the gold standard for establishing causality 

• The poor alternative to experimentation

  • Company leaders pick policies based on intuition

• Random assignment is essential

  • Ensures potential confounds are balanced across groups

• Larger sample sizes are more likely to reveal truth about population but not always possible (time or money constraints)

Limitations of experiments

1. Expensive to run

2. Infeasible for large, systemic changes 

3. Short run experiments vs longer term effects 

4. Incremental improvements don’t always lead to a global optimum 

Causal Inferences from Observational Data

• Statistics and econometrics can help us draw causal inferences from observation data without the need for an experiment

• Examples 

  • Natural experiments 

  • Instrumental variables

    • Only some of the variation in X is random 

Causal Inference Doesn't Tell You Everything

• If you manipulate an IV, what else will happen? 

• The ideal sequence

  • Develop insights and hypotheses via descriptive methods ←→

  • ←→Test hypotheses via experimentation or causal inference methods